Automated emotion identification via physiological data from wearable devices is a growing field, yet traditional electroencephalography (EEG) and photoplethysmography (PPG) collection methods can be uncomfortable. This research introduces a novel structure of the in-ear wearable device that captures both PPG and EEG signals to enhance user comfort for emotion recognition. Data were collected from 21 individuals experiencing four emotional states (fear, happy, calm, sad) induced by video stimuli. Following signal preprocessing, temporal and frequency domain features were extracted and selected using the ReliefF approach. Classification accuracy was assessed for PPG, EEG, and combined features, with combined features yielding superior results. An XGBoost classifier, optimized with Bayesian hyperparameter tuning, achieved 97.58% accuracy, 97.57% precision, 97.57% recall, and a 97.58% F1 score, outperforming support vector machine, decision tree, random forest, and K-Nearest Neighbor classifiers. These findings highlight the benefits of multimodal physiological sensing and optimized machine learning for reliable emotion characterization, with implications for mental health monitoring and human-computer interaction.

Over the past decades, China has faced an increasingly severe fire issue, resulting in significant human and financial losses. Consequently, the importance of enhancing fire safety education for citizens has become more pronounced. However, due to safety risks in actual fire scenes, the practical use of firefighting equipment was often learned passively through text or video resources, posing challenges in ensuring users can proficiently operate actual equipment. Virtual reality (VR) emerged as a promising technology to provide users with a low-risk environment to gain hands-on experience with firefighting equipment. Therefore, this study aimed to assess the effectiveness of a specially designed VR game-based firefighting extinguishing equipment training (FEET) that integrated VR with serious gaming elements. The study evaluated the effectiveness of FEET through a comparison between high (HI) and low (LI) immersiveness VR and a non-immersive (NI) setting as a baseline. The study incorporated haptic feedback mechanisms, such as audio and vibration, to enhance user immersiveness within the HI setting. The focus of the study was on evaluating knowledge acquisition, user experiences, and interaction preferences for the proposed FEET design. The study adopted three distinct learning sequences to mitigate the VR halo effects. The results of the study indicated that both VR-based learning methods yielded outstanding user experiences that significantly surpassed the NI setting. Furthermore, the study showed that haptic feedback with richer game contents further enhanced user experiences, underscoring the importance of simulating realistic sensing in improving learning outcomes.

Monitoring mental states is essential for understanding physiological and neurological health, offering insights that enhance diagnostics and therapy. However, current systems are often bulky, immobile, and reliant on external devices for deep learning (DL), limiting real-time applications. This study introduces a compact, real-time mental state monitoring system with four innovations. (1) First, a wearable ear-centric device captures multimodal biosignals—behind-the-ear (BTE) electroencephalography (EEG) and in-ear photoplethysmography (PPG)—enhancing portability. (2) Second, a high-resolution superlet transform (HRST) converts these signals into spectrogram images to enrich input features for DL models. (3) Third, the PhysAttenConvLSTM model, designed for edge deployment using Tiny Machine Learning (TinyML), integrates convolutional neural networks (CNNs), long short-term memory (LSTM), self-attention, and a physics-informed merge layer (FIML) to classify affective states efficiently. All processes—from signal acquisition to classification—occur directly on the edge device, enabling low-latency, real-time operation without external computation. (4) Fourth, an Internet of Medical Things (IoMT) framework is implemented for data storage, remote access, and system management. Multimodal data were collected from 25 participants during social media-based mental state induction tasks. The system achieved 92.07% accuracy in leave-one-out cross-validation (LOOCV) and 95.05% in 10-fold cross-validation (CV), with strong specificity, sensitivity, precision, and F1 scores. This research advances edge DL, multimodal biosignal fusion, and IoMT integration, offering an ultra-portable, energy-efficient solution for real-time mental state monitoring, with potential in affective computing and personalized mental health care.

Machine learning (ML) techniques are applied for profiling computing and processing resources data collected while running deep neural network models on edge devices. Adaptive deep neural network (DNN) model switching requires proper benchmarking for categorizing AI models based on their applications and computational resources enabled by their processing accelerators. Based on benchmark metrics, DNN models can be classified into tiny, low, small, medium, and large resources, then identify DNN models that perform well within resource constraints. Ensure efficient resource allocation, latency management, and trade-off between accuracy and resource. In this work, we propose a benchmark for edge transfer artificial intelligence learning service (TALS) that uses ML techniques. They aim at classifying DNN models by their target edge applications while running edge inferences. We used both unsupervised learning (UL) and supervised learning (SL) techniques to identify the most effective features for the TALS models and to benchmark the performance of edge devices. To achieve this, two approaches were investigated: first, determining features based on edge inference’s computing resources profiling using principal component analysis; and second, classifying the DNN models at the target application level using a regression approach based on historical resource utilization data. In addition, we propose a dynamic model transfer learning that switches between a set of pre-trained and optimized and quantized DNN models based on the cost function. ML techniques learn resource-aware prediction from new resource allocation data and ensure that the multicriteria switch cost selects the inference task models that meet the edge resource constraint requirements. The experimental results highlight a strong relationship between the supervised learning model and the clustering execution method. The dynamic switching approach on real edge devices demonstrates dynamic switching between models according to inference task complexity. We conclude that dynamic switching models allows to ensure smooth operation without overloading resources in edge intelligence.

The deployment and maintenance of the underwater sensor network are cumbersome and require a significant amount of manual labor, resulting in reduced efficiency and an increased likelihood of errors. This study presents a compact underwater wireless sensor network that use multiple diversity gain-enabled relays to receive and transmit data from sensor nodes to the gateway. The relay is powered by employing combining techniques, such as equal gain combining (EGC), majority logic combining (MLC), and selection combining (SC) to augment its performance. The sensor node is fitted with an underwater optical wireless communication (UWOC) module, which wirelessly transfers the inertial measurement unit (IMU) sensor data to the nearest relay. The EGC, MLC, and SC have achieved packet error rates of 26%, 28%, and 30%, respectively, at a transmission rate of 0.5 Mb/s. The accelerometer and gyroscope data collected from multiple fish, including the sensor readings of roll, pitch, yaw, speed, and dynamic acceleration, are utilized to train long short-term memory (LSTM), spatial attention, recurrent neural network (RNN), Transformer, and gated recurrent unit (GRU) machine learning models. The models are capable of predicting the particular fish’s steady, low, medium, and high acceleration states. The LSTM, spatial attention, RNN, Transformer, and GRU ML models achieved training accuracy rates of 79.06%, 93.86%, 91.07%, 87.38%, and 68.16%, respectively.

The study introduces an innovative approach to efficient mental stress detection by combining electroencephalography (EEG) analysis with on-chip neural networks, taking advantage of EEG's temporal resolution and the computational capabilities of embedded neural networks. The proposed system utilizes behind-the-ear (BTE) EEG signals and on-chip neural networks for mental stress detection. A wearable custom-designed device captures EEG signals from a single BTE channel, performs on-chip signal-to-spectrogram conversion, and integrates a compact convolutional neural network (CNN) for stress classification. The system systematically identifies key EEG frequency bands associated with stress and includes a user-friendly smartphone application for intuitive stress monitoring. EEG data were collected from 15 participants during stress-inducing tasks, such as Stroop and Mental Arithmetic tests. On-chip processing is essential for filtering EEG noise, converting signals into spectrogram images, and using these images as inputs for stress detection through the proposed on-chip CNN model. The experimental results demonstrate strong performance: leave-one-out cross-validation (LOOCV) achieves 91.72% accuracy, 93.74% specificity, 88.69% sensitivity, 90.43% precision, and an F1-score of 0.8955; while 10-fold cross-validation (CV) yields 95.32% accuracy, 95.89% specificity, 94.47% sensitivity, 93.95% precision, and an F1-score of 0.9421 on untrained datasets. The Beta band (13 Hz to 30 Hz) is identified as the most significant frequency band for detecting mental stress. This integration of BTE EEG analysis with on-chip CNNs represents a significant advancement in mental stress detection and has potential applications in medical assistance tools.

Integrating computer vision and physiological analysis in driver drowsiness detection (DDD) is a promising technology for accurately identifying drowsy states while driving, thereby preventing potentially dangerous accidents. This study proposes a multimodal DDD system with a deep neural network that combines computer vision-based face expression analysis and electroencephalogram (EEG) data analysis. Key contributions include: 1) providing a comprehensive hardware, firmware, and software design for the DDD system to acquire behind-the-ear (BTE) EEG signals, rather than conventional scalp EEGs, due to their convenience and practicality; 2) proposing a powerful and lightweight GAN-based auto-denoising method to eliminate artifacts from EEG signals during signal acquisition, significantly influencing the quality of the obtained result; 3) developing a multimodal DDD network by combining EEG analysis and computer vision-based face expression identification to improve performance in monitoring and early detection of the driver’s drowsiness while engaging in traffic. The study employs the relative root mean squared error (RRMSE) in both temporal and spectral domains to quantitatively assess the performance of the proposed approaches in artifact removal. The proposed GAN-based auto-denoising network outperforms other comparable approaches, with an RRMSE (temporal) of 0.210 and RRMSE (spectral) of 0.161. The proposed trained multimodal model with GAN-based auto-denoising is superior to other models with different denoising approaches in driver drowsiness detection across all five-evaluation metrics, with an accuracy of 95.33%, specificity of 95.48%, sensitivity of 95.17%, precision of 95.47%, and an F1-score of 95.32%. The experimental results demonstrate the practicality and feasibility of our proposed DDD system.

Abstract This study proposes a non-invasive method to estimate both color and sensory attributes of Shine Muscat grapes from standard camera images. First, we focus on color estimation by integrating a Vision Transformer (ViT) feature extractor with interquartile range (IQR)-based outlier removal. Experimental results show that our approach achieves 97.2% accuracy, significantly outperforming Convolutional Neural Network (CNN) models. This improvement underscores the importance of capturing global contextual information to differentiate subtle color variations in grape ripeness. Second, we address human sensory evaluation by collecting questionnaire responses on 13 attributes (e.g., “Sweetness,” “Overall taste rating”), each rated on a five-point scale. Because these ratings tend to cluster around midrange values (labels “2,” “3,” and “4”), we initially limit the dataset to the extreme labels “1” (“lowest grade”) and “5” (“highest grade”) for binary classification. Three attributes—“Overall color,” “Sweetness,” and “Overall taste rating”—exhibit relatively high classification accuracies of 79.9%, 75.1%, and 75.7%, respectively. By contrast, the other 10 attributes reach only 50%–66%, suggesting that subjective variations and limited visual cues pose significant challenges. Overall, the proposed approach demonstrates the feasibility of an image-based system that integrates color estimation and sensory evaluation to support more objective, data-driven harvest timing decisions for Shine Muscat grapes.

Wearable electroencephalography (EEG) devices are emerging as crucial tools in human-robot interaction (HRI), enabling intuitive and effective communication between humans and robots. These devices non-invasively measure brain activity, providing real-time insights into a user’s mental state, intentions, and cognitive load. This paper explores the advancements and applications of wearable ear EEG technology in HRI, with a focus on detecting human intent, monitoring emotional and cognitive states, and delivering real-time feedback for adaptive robot behavior. The benefits of wearable EEG over traditional scalp EEG, such as enhanced user comfort, reduced setup time, and improved long-term wearability, are thoroughly examined. Additionally, the paper covers advanced applications, including emotion-aware adaptive interactions, neurofeedback training, and direct robot control via brain-computer interfaces (BCIs). A review of the latest advancements in device miniaturization, integration with other wearable sensors, and applications in virtual reality, gaming, and healthcare is provided. Future directions focus on expanding applications, enhancing human-AI collaboration, and improving accuracy and reliability in HRI. The studies discussed in this paper represent our state-of-the-art research across various fields and applications of wearable EEG systems. This study highlights the potential of wearable Ear-EEG devices to revolutionize human-robot interactions by enhancing customization, responsiveness, and overall efficacy.

This study identifies challenges in user interface design for applying Augmented Reality (AR) technology to support work for individuals with intellectual disabilities. The main focus is on difficulty in self-assessing fatigue levels. The research methodology involved conducting experiments using HoloLens 2, analyzing changes in biometric information during fatigue, and examining the relationship between information display position and head orientation. Results indicate that changes in head height and orientation could potentially serve as fatigue indicators. In conclusion, while AR technology is effective in supporting work for individuals with intellectual disabilities, special considerations are necessary. Fatigue detection using biometric information and optimization of information display positions are crucial, and these findings may lead to safer and less burdensome use of AR.

Infrared thermal imaging presents significant potential in various domains. However, the widespread development of this technology is hindered by the high cost associated with acquiring high-quality thermal imaging sensors. To overcome this challenge, super-resolution techniques have emerged as a viable solution for extracting valuable information from low-resolution thermal images. While generative adversarial networks (GANs) have been widely adopted for thermal imaging super-resolution, their performance is limited by the inherent lack of detail in low-resolution training images, resulting in reduced fidelity and accuracy in generating high-resolution reconstructions. To tackle this challenge, this letter introduces FEGAN, a novel approach that enhances the performance of GANs by incorporating a feature-oriented enhanced (FE) mechanism within the generative network (GN). The FE plays a pivotal role in extracting high-frequency texture and edge details from lowresolution inputs and reconstructing them into enhanced images. This process substantially improves textures and edges within the training set of thermal images. Furthermore, refinements have been applied to both the GN and the discriminative network (DN) to enhance feature extraction efficiency. The experimental findings unequivocally demonstrate the superior performance of FEGAN compared to state-of-the-art methods. FEGAN achieves impressive performance metrics, including PSNR of 27.18, SSIM of 0.6523, FSIM of 0.5500, and LPIPS of 0.1221, highlighting its remarkable capabilities in the realm of thermal image superresolution.

Indoor positioning in the firefighting ground shows promising application prospects for enhancing rescue safety and efficiency. Low visibility and signal interference caused by smoke pose significant difficulties for visual Simultaneous Localization and Mapping (SLAM) systems and radio frequency-based localization methods, while the performance of existing pedestrian dead reckoning (PDR) methods is affected by unpredictable user gaits. This paper introduces a gait analysis-based PDR (GA-PDR) for deriving heading estimation in PDR to improve its localization performance. The proposed method determines the step pattern by analyzing the features of inertial measurement unit data, thereby enabling the classification of forward, left- and right-turn and around-turn from left or right-side movements. In addition, this study introduces a redundant turn elimination method to differentiate false positive patterns via a time-domain heading analysis for turn movements. The location tracking performance with the proposed GA-PDR approach is validated using a self-created dataset established under a smoke-filled experiment, the results of which indicate a lower loop closure error compared with the traditional PDR in all of the tested scenarios.

For Internet of Things (IoT) connectivity, visible light communication (VLC) can play an important role, compared to traditional radio frequency (RF) communication. VLC does not interfere with existing RF communication, and the frequency spectrum is unregulated. Additionally, in an indoor area, VLC can provide secure communication. This study proposes an authentication-based framework that enhances security at the user end. In addition, we have modified the multiple pulse position modulation (MPPM) to map digits and send particulate matter (PM) data in broad area coverage. To reduce communication channel usage, we take the difference between two consecutive values instead of sending the whole value, which enables us to transmit the same volume of data in fewer optical signals using the proposed modulation. The comparative analysis result shows that the proposed modulation can transmit PM data more efficiently than other modulation techniques. We also provided delay analysis for proper synchronization for connectivity between nodes. The modulation was tested in the case of multihop connectivity to send data on long distance about 36 m in real time.

Visible light communication (VLC) exhibits great potential in connecting Internet of Things (IoT) devices. The connection of a huge number of IoT devices by utilizing the existing radio-frequency spectrum is a challenging task. Therefore, a new frequency spectrum is required for a smooth operation. In this work, we investigate a multiple-input multiple-output (MIMO) VLC system capable of connecting IoT devices to achieve long-range indoor communication. To monitor the indoor environment, a monitoring system for collecting and transmitting particulate matter, temperature, and humidity data using MIMO VLC is proposed. Four different single-carrier techniques are tested, and their error performance is analyzed by experimental trials. Next, the maximum communication range, which is important for efficient network planning and uninterrupted connectivity, is evaluated. Two antenna configurations (2$\times $2 and 4$\times $4) are tested for MIMO VLC. In the 2$\times $2 configuration, a 14.5–21 m transmission distance is achieved by employing four different modulation techniques. In the 4$\times $4 communication, a 7–10 m transmission distance with a reliable error rate is achieved by employing three different modulation techniques. In addition, a lightweight encryption technique is used to enhance data security. The proposed system provides an efficient solution for monitoring different types of data using indoor IoT connectivity.

Edge intelligence (EI) refers to a set of connected systems and devices for artificial intelligence (AI) data collected and learned near the data collection site. The EI model inference phase has been improved through edge caching technologies such as intelligent models (IMs). IM inference across heterogeneously distributed edge nodes is worthy of discussion. The present focuses on software-defined infrastructure (SDI) and introduces a containerized EI framework for a mobile wearable Internet-of-Things (IoT) system. This framework, called the containerized edge intelligence framework (CEIF), is an inter-working architecture that allows the provisioning of containerized EI processing intelligent services related to mobile wearable IoT systems. CEIF enables dynamic instantiation of the inference services of AI models that have been pre-trained on clouds. It also accommodates edge computing devices (ECDs) running the container virtualization technique. Dynamic AI learning policies can also help with workload optimization, thereby reducing the response time of the requests of the EI inference. To stall the rapid increase in user workload when inferring the collected data for analysis, we then propose a deep q-learning algorithm in which the container cluster platform learns the varying user workload at the location of each ECD. The requests of the EI inference are scaled with the learned value and are processed successfully without overloading the ECD. When evaluated in a case study, the proposed algorithm enabled scaling of the processing requests of the EI inference in a containerized EI system while minimizing the number of instantiated container EI instances. The EI inference's requests are completed in an under-loaded container EI cluster system.

These particulate matter (PM) causes lethal diseases to humans, and both short term and long term exposure are known to have hazardous effects. PM10and PM2.5have diameters less 10 and 2.5 μm, respectively, which makes them more dangerous to the human body. Thus, information regarding PM concentrations in indoor environments of dust-sensitive places is essential for proper precaution. In this study, we measured PM values in a large area and transferred this information using visible light communication (VLC) to the monitoring node. VLC is considered as an efficient technique due to its unique advantages and is currently unregulated. We applied bidirectional VLC to transfer information to ensure two-way communication. We also applied a multihop strategy to make the system function as a query answering system at extended distances. At one end, we generated a request to relay the request to the proper node and received the response in the form of PM data. In our experiment, we implemented four nodes and conducted multihop communication utilizing only a VLC link. We achieved a distance of 13.5 m with a zero-error rate between the two nodes using nonreturn to zero on-off keying (NRZ-OOK) modulation. In case of multihop, we achieved a distance greater than 40 m using four nodes to send dust information with a minimal error rate. This can be applied to large indoor areas where radio frequency is restricted.

In this paper, we present the design and implementation of a wireless, wearable brain machine interface (BMI) system dedicated to signal sensing and processing for driver drowsiness detection (DDD). Owing to the importance of driver drowsiness and the possibility for brainwaves-based DDD, many electroencephalogram (EEG)-based approaches have been proposed. However, few studies focus on the early detection of driver drowsiness and on the early management of driver drowsiness using a closed-loop algorithm. The reported wireless and wearable BMI system is used for 1) simultaneous EEG and gyroscope-based head movement measurement for the early detection of driver drowsiness and 2) simultaneous EEG and transcranial direct current stimulation (tDCS) for the early management of driver drowsiness. To achieve the purposes of easy-to-use and distraction-free driving, a Bluetooth low-energy module is embedded in this BMI system and used to communicate with a fully wearable consumer device, a smartwatch, which coordinates the work of drowsiness monitoring and brain stimulation with its embedded closed-loop algorithm. The proposed system offers a 128 Hz sampling rate per channel, 12-bit and 16-bit resolution for a single-channel EEG and a three-channel gyroscope, and a maximum 2 mA current for the tDCS. The current consumption of the whole headset system is 56 mA. The battery life of the smartwatch is 9 h. The DDD experimental results show that the proposed system obtained a 93.67% five-level overall accuracy, a 96.15% two-level (alert versus slightly drowsy) accuracy, and maximum 16- to 23-min wakefulness maintenance.

Increased driver stress is generally recognized as one of the major factors leading to road accidents and loss of life. Even though physiological signals are reported as the most reliable means to measure driver stresses, they often require the use of unique and expensive sensors, which produce dynamic and varying readings within individuals. This paper presents a novel means to predict a driver's stress level by evaluating the movement pattern of the steering wheel. This is accomplished by using an inertial motion unit sensor, which is placed on a glove worn by the driver. The motion sensor selected for this paper was chosen because for its low cost and the fact that it is least affected by environmental factors as compared with a physiological signal. Experiments were conducted in three different environmental scenarios. The scenarios were classified as “urban,” “highway,” and “rural,” and they were chosen to simulate contrasting stress conditions experienced by the driver. In this paper, skin conductance and driver self-reports served as a reference stress to predict the driver's stress level. Galvanic skin response, a well-known stress indicator, was captured along the driver's palm and the readings were transmitted to a mobile device via low energy Bluetooth for further processing. The results revealed that indirect measurement of steering wheel movement with an inertial motion sensor could obtain accuracies up to an average rate of 94.78%. This demonstrates the opportunity for inclusion of motion sensors in wireless driver assistance systems for ambulatory monitoring of stress levels.

For embodying and implementing the ubiquitous computing environments, the recent increase in concern for communication systems has been leading the evolution of the theory and technology of the related fields. Among these network-related problems, the minimum spanning tree (MST) problems have many applications in the communication systems field, because they are solved as one of the conventional combinatorial optimization problems. In particular, the optimal communication spanning tree (OCST) problem is a famous application of the MST problem. The OCST problem is defined by finding a spanning tree that connects all nodes and satisfies their communication requirements for connecting all nodes. This OCST problem can be applied to many network optimization fields, such as network topology design, multicast tree configuration, and ad-hoc network or ubiquitous sensor network routing, etc. This paper presents a genetic algorithm with a new encoding method, which is based on the Prüfer number (PN) and a clustering string, for solving the OCST problems. We will also demonstrate that the efficiency and effectiveness of our proposed method can be shown by several experimental results, employing the proposed method as the solution method of the OCST problems.